San Francisco: The global debate over artificial intelligence safety and cybersecurity has taken a significant turn as Nvidia, Microsoft, SpaceX, Palantir and dozens of other technology companies launched the Open Secure AI Alliance (OSAA), a collaborative initiative aimed at strengthening the security of open AI models. The announcement comes in the wake of a high-profile cyberattack involving rogue OpenAI models, an incident that has intensified discussions around the role of open-source AI in cyber defence and national security.
According to Nvidia, the alliance will focus on identifying, remediating and publicly disclosing vulnerabilities using open technologies. The company said the initiative seeks to improve the security and resilience of AI systems by enabling organisations to inspect, modify and deploy advanced AI models on their own infrastructure.
The move follows revelations surrounding a recent cyberattack targeting AI startup Hugging Face. Reports indicated that the company struggled to defend itself using leading proprietary U.S. frontier AI models because built-in safety guardrails were unable to distinguish between malicious attackers and legitimate defenders. Instead, Hugging Face reportedly relied on a self-hosted, open-weight Chinese AI model, which did not impose the same operational restrictions and proved more effective in supporting cyber defence efforts.
Nvidia described the incident as a clear demonstration of the importance of open AI systems for cybersecurity. The company argued that organisations need access to frontier-level AI that can be independently inspected and adapted, especially during critical security incidents where rapid response is essential.
The Open Secure AI Alliance includes several major technology firms from the United States and Europe, including Nvidia, Microsoft, SpaceX and Palantir. Together, the members aim to develop and share open AI tools that enhance cybersecurity while promoting transparency and collaboration within the AI ecosystem.
The initiative also arrives amid growing political scrutiny in Washington over the rapid adoption of Chinese-developed AI models. Many of the world’s most capable open-weight models originate from Chinese companies, prompting concerns among U.S. policymakers about intellectual property theft and national security.
Chinese AI firms have faced allegations of conducting “distillation” campaigns, a technique in which one AI model extracts knowledge from another, more advanced model. U.S. Treasury Secretary Scott Bessent recently warned that Chinese companies engaging in such practices could face sanctions, signalling a tougher stance on AI-related intellectual property violations.
Chris McGuire, Senior Fellow for China and Emerging Technologies at the Council on Foreign Relations, noted that Washington’s discussions are less about the open-source versus closed-source debate and more about addressing concerns over alleged Chinese intellectual property theft. He suggested that potential government actions could include restrictions on commercial transactions involving Chinese AI models, such as purchasing API access or hosting the models on cloud platforms for commercial inference services.
However, industry leaders have cautioned against imposing sweeping restrictions on open-weight AI. Last week, Nvidia, Microsoft, Meta, Palantir and more than 20 other technology companies jointly urged policymakers to avoid what they described as “premature restrictions” that could undermine innovation, reduce competition and encourage AI development to move overseas.
Supporters of open AI argue that allowing developers and organisations to inspect and self-host advanced models improves transparency, strengthens cybersecurity and enables faster responses during emergencies. They contend that limiting access to such technologies could inadvertently weaken cyber defences by preventing security teams from deploying adaptable AI tools when they are needed most.
As governments worldwide continue to formulate AI regulations, the launch of the Open Secure AI Alliance underscores the technology industry’s growing emphasis on balancing innovation, security and openness.
